1 Fall 2006 – 7.03

7.03, 2006, Lecture 23 Eukaryotic Genes and Genomes IV

In the last three lectures we have thought a lot about analyzing a regulatory system in S. cerevisiae, namely Gal regulation that involved a hand full of genes. These studies monitored the increased transcription of Gal genes in the presence of galactose (and the absence of glucose); we saw that this regulation is achieved by particular proteins, or multiprotein complexes that bind to specific sequences in the promoter region upstream from their target genes. What if I told you that it is now possible to do the following in S. cerevisiae:

! Monitor mRNA expression level for every gene in S. cerevisiae, in one single experiment.

! Monitor all the binding sites in the S. cerevisiae genome for each transcription factor in a single experiment.

! Determine all possible pair-wise interactions for every S. cerevisiae protein.

Obviously I wouldn’t mention these possibilities if they weren’t already happening. What I want to do today is to introduce you to the idea of carrying out genetic analyses on a global, genome-wide scale, and hopefully give you some examples that are relevant to what we have already learned along the way. So, this will be

a technology oriented lecture, but with some application to what we have already learned about gene regulation in eukaryotes. It should also be mentioned that what will be described for S. cerevisiae, is theoretically possible for any organism whose genome has been completely sequenced and the location of all the genes in that genome have been established. So, what we will learn today is being, or will be, applied to higher eukaryotes and mammals.

! Monitor mRNA expression level for every gene in S. cerevisiae, in

one single experiment: Global transcriptional profiling. Before we consider how it is possible to measure the levels of thousands of mRNA species, we will have to step back to consider how the levels of one or two mRNA species can be measured by Northern Blot analysis….and I know you must have learned this in 7.01 if not in high school. Northern blot analysis is based upon the fact that DNA and RNA molecules that possess complementary base sequences will hybridize together to form a double stranded molecule. If the complementarity is perfect the duplex molecule is stable, if it is imperfect (with base pair mismatches)

Numbers of genes per haploid genome

5,800

14,000

19,000 22,500

22,500

Numbers of genes per haploid genome

5,800

14,000

19,000 22,500

22,500

2 Fall 2006 – 7.03

it is relatively less stable. This provides the specificity needed to identify perfectly matched DNA:RNA duplexes (on Northern Blots) and DNA:DNA duplexes (on Southern Blots). This specificity is needed to be sure we are measuring the level of one particular transcript and that this is not contaminated with signal from closely related transcripts. RNA is isolated from cells, size fractionated on a gel;

the thousands of mRNAs species form a smear on the gel which is punctuated by the strong ribosomal RNA bands (28S and 18S) that do not interfere with the analysis.

The breakthrough in developing microarrays for analyzing mRNA levels was to reverse the logic – instead of immobilizing the mRNAs for hybridization with one or two labeled complementary DNA (cDNA) probes, all possible cDNA probes are immobilized on a solid surface (usually glass slides). The spoting of probes is achieved robotically; the DNA probes are designed to specifically hybridize to only one nucleic acid sequence that represents a single mRNA species. The thousands of DNA probes are dispensed from 96-well, or 384-well plates to an addressable site on the solid surface. The mRNA population from each cell type purified and then copied such that the copy is fluorescently labeled. This fluorescent population is hybridized to the immobilized probes, and the intensity of the fluorescence at each probe spot is proportional to the number of copies of that specific mRNA species in the original mRNA population.

!"# %&&'(%l%*#+ &,-. /'/0la2%'3 %4 /5'(#+ 6"7(5%+%*#+89%2" :;0#3c#4 4/#c%@%c @'5 '3# '5 29' A#3#

/5'+0c24

All mRNAs separated by

size

!"# %&&'(%l%*#+ &,-. /'/0la2%'3 %4 /5'(#+ 6"7(5%+%*#+89%2" :;0#3c#4 4/#c%@%c @'5 '3# '5 29' A#3#

/5'+0c24

All mRNAs separated by

size

-'52"#53 Bl'24Immobilized mRNA population hybridized with labeled DNA probe representing one

or two genes

C-. D%c5'a55a74Immobilized DNA probes representing all

possible geneshybridized with labeled mRNA population

3 Fall 2006 – 7.03

So let’s look at how this would actually work in a real experiment. mRNA is isolated from yeast cells in state A (e.g., minus galactose) and from yeast cells in state B (e.g., plus galactose), and copies of each population is made such that one fluoresces red and the other fluoresces green. After mixing, these fluorescent molecules are hybridized to the slides containing ~5,800 DNA probes, each one specific for detecting hybridization of many copies of an individual mRNA species.

The location and identity of each probe on the microarray slide is known, and each probe is specific for a single mRNA. The color and intensity of the fluorescence is measured by scanning the slide with lasers, and the relative abundance of each mRNA in the cells of State A vs State B can be calculated from the emitted fluorescence, i.e., the relative level of 5,800 mRNAs can be compared between two populations of yeast cells.

Presenting data for thousands of mRNA transcripts is clearly a challenge. You could present endless tables of data, but our brains are much more adept at

recognizing shapes, patterns and colors. Colored representations of up and down regulation of transcripts levels is the preferred way to present data.

Yeast in state A

AAAAAAAAAA

AAAAA

AAAAA

TTTTTTTTTT

TTTTT

TTTTT

Yeast in state B

AAAAAAAAAAAAAAA

AAAAA

TTTTTTTTTT

TTTTT

TTTTT

La(#l copies of mRNA species with RED or GREEN

Isolate mRNA populations

MIX

TTTTT

TTTTT

TTTTT

TTTTT TTTTT

TTTTTTTTTT

TTTTTF7(5%+%*# to the

microarray

Yeast in state A

AAAAAAAAAA

AAAAA

AAAAA

TTTTTTTTTT

TTTTT

TTTTT

Yeast in state A

AAAAAAAAAA

AAAAA

AAAAA

AAAAAAAAAA

AAAAA

AAAAA

TTTTTTTTTT

TTTTT

TTTTT

TTTTTTTTTT

TTTTT

TTTTT

Yeast in state B

AAAAAAAAAAAAAAA

AAAAA

AAAAAAAAAAAAAAA

AAAAA

AAAAAAAAAA

AAAAA

TTTTTTTTTT

TTTTT

TTTTTTTTTTTTTTT

TTTTT

TTTTTTTTTTTTTTT

TTTTT

TTTTT

La(#l copies of mRNA species with RED or GREEN

Isolate mRNA populations

MIX

TTTTTTTTTT

TTTTT

TTTTT

TTTTT TTTTT

TTTTTTTTTT

TTTTTTTTTTTTTTT

TTTTTTTTTT

TTTTTTTTTT TTTTTTTTTT

TTTTTTTTTTTTTTTTTTTT

TTTTTTTTTTF7(5%+%*# to the

microarray

G"a2H4 "a//#3%3A a2 #ac" 4/'2I

Hybridization

G"a2H4 "a//#3%3A a2 #ac" 4/'2I

Hybridization

-'52"#53 Bl'2 J4K D%c5'a55a7-'52"#53 Bl'2 J4K D%c5'a55a7

Each colored vertical line in the horizontal lane displays the relative expression level of a single mRNA

Each colored vertical line in the horizontal lane displays the relative expression level of a single mRNA

mRNA present much higher in State A than State B

mRNA present much higher in State B than State A

mRNA present at equal levels in States A and B

mRNA present much higher in State A than State B

mRNA present much higher in State B than State A

mRNA present at equal levels in States A and B

4 Fall 2006 – 7.03

For our purposes here, lets look at what genes are up-regulated when a glucose grown culture of S. cerevisiae is shifted into galactose; what genes are up-regulated under these conditions? Obviously transcripts for Gal1, Gal7 and Gal10 genes will be up-regulated, as we have discussed in the last couple of

lectures. But upon looking globally, it is now clear that some other genes are also up-regulated. In addition, Gal2 (galactose permease) and Gal80 (the negative regulator of the Gal4 Transcriptional activator) are also induced; this was previously known, although we didn’t discuss it diretly in the previous lectures. but upon looking globally, it has become clear that some other genes are also up-regulated. (This figure shows just a small snapshot of the response.) These additional genes are Fur4, Gcy1, Mth1, and

Pcl10, and their co-regulation along with the Gal genes was previously unrealized. We will be coming back to this later in the lecture.

! Monitor all the binding sites in the S. cerevisiae genome for each transcription factor in a single experiment.

In the last lecture we talked about deletion analysis of cis-acting regulatory sequences identifying the location of UAS and URS sequences upstream of the Gal1 gene. That the Gal4 transcriptional activator protein binds to the DNA sequence present at the URSGAL1 can be shown to happen in the test tube, but showing that it is actually bound in a living cell is another matter. A method was recently developed for doing just that, and this method has been further

What transcripts have increased levels when shifted from glucose to galactose?

Lc%#3c#M N'l ;OPM Q440# RRPPM ;:PS-;:PO M ;; C#c#&(#5 ;PPP

What transcripts have increased levels when shifted from glucose to galactose?

Lc%#3c#M N'l ;OPM Q440# RRPPM ;:PS-;:PO M ;; C#c#&(#5 ;PPP

Chromatin Immuno Precipitation (ChIP)

Formaldehyde treatment crosslinks

proteins to DNA

Isolate DNA with proteins crosslinked,

shear into small fragments

Immunoprecipitatespecific transcription factor and its bound

DNA

Reverse the formaldehyde

crosslinks and get rid of protein

DNA fragments that the transcription

factor was bound to in the living yeast cell

L%J%3A c#ll4H2CO

Chromatin Immuno Precipitation (ChIP)

Formaldehyde treatment crosslinks

proteins to DNA

Isolate DNA with proteins crosslinked,

shear into small fragments

Immunoprecipitatespecific transcription factor and its bound

DNA

Reverse the formaldehyde

crosslinks and get rid of protein

DNA fragments that the transcription

factor was bound to in the living yeast cell

L%J%3A c#ll4H2CO

Nature Reviews Immunology 4; 381-386 (2004);

Label this enriched DNA with 5#+ fluorescence

Grossly #35%c"#+ for DNA that was bound by the purple transcription

factor

Mix with 03-#35%c"#+genomic DNA labeled

with A5##3 fluorescenceScience, Vol 290, Issue 5500, 2306-2309 , 22 December 2000

Nature Reviews Immunology 4; 381-386 (2004);

Label this enriched DNA with 5#+ fluorescence

Grossly #35%c"#+ for DNA that was bound by the purple transcription

factor

Mix with 03-#35%c"#+genomic DNA labeled

with A5##3 fluorescenceScience, Vol 290, Issue 5500, 2306-2309 , 22 December 2000

5 Fall 2006 – 7.03

developed to determine transcription regulator binding across the whole genome. This method takes advantage of the fact that formaldehyde crosslinks proteins to DNA in a way that can later be reversed. For galactose grown yeast cells chromatin immunoprecipitation (ChIP) with an

antibody that pulls down the Gal4 protein revealed some surprises. In addition to confirming that Gal4 binds to the promoters regions upstream of the expected Gal genes, the Gal4 protein also binds to the promoter regions of 4 other genes, namely Fur4, Pcl10, Mth1 (shown in the adjacent figure) and Gcy1 (not shown). Note that these genes were shown to be induced by galactos in the previous section. Just how the up-regulation of Fur4, Pcl10 and Mth1 might contribute to optimizing the metabolism of galactose is shown in this figure, but

the role Gcy1 plays is unclear. Clearly, taking a global look at what genes are up-regulated in the presence of galactose, and taking a global look at what promoters are bound by the Gal4 regulator, has clearly enriched our view of how S. cerevisiae adapts to the presence of this sugar. The ChIP approach, followed by hybridization to DNA microarrays, was originally limited to monitoring binding of transcriptional regulators for which there were good precipitating antibodies. However, this limitation was recently eliminated by fusing an epitope TAG to each regulator gene. This epitope TAG is recognized by a strong antibody, and so a single antibody can “pull down” (immunoprecipitate) >100 different regulatory proteins, each of which is expressed in its own yeast strain.

Gal: Gal8PGal: Gal8P

Gal: Gal8PGal: Gal8P

Gal: Gal8PGal: Gal8P

HXT is a transporter for hexose/glucose

A more complete view of galactoseinduced gene expression in S. cerevisiae

Gal: Gal8PGal: Gal8P

Gal: Gal8PGal: Gal8P

Gal: Gal8PGal: Gal8P

HXT is a transporter for hexose/glucose

A more complete view of galactoseinduced gene expression in S. cerevisiae

Science, Vol 290, Issue 5500, 2306-2309 , 22 December 2000

.55a7#+ /5'(# 4#>0#3c#4 5#/5#4#32 2"# 0/425#a& c%4-

ac2%3A 5#A%'34 '@ all RM8PP A#3#4

Science, Vol 290, Issue 5500, 2306-2309 , 22 December 2000

.55a7#+ /5'(# 4#>0#3c#4 5#/5#4#32 2"# 0/425#a& c%4-

ac2%3A 5#A%'34 '@ all RM8PP A#3#4

,#A0la2'57

6 Fall 2006 – 7.03

This has enabled a massive study to identify all of the target genes for each of 106 transcriptional regulators in S. cerevisiae growing in a defined medium. A compilation of all the data has revealed a number of fundamentally different regulatory motifs; these are shown in the adjacent figure. For the most part the Gal4 regulatory network (now shown) represents a simple Single Input Motif. This approach has already been extended to human cells and it will not be long until detailed regulatory mechanisms are defined for humans, in the way it is now happening in yeast. It is now possible to go on to monitor which genes the transcriptional regulators bind to under different environmental conditions, and from there to build more dynamic models for how these genetic regulatory mechanisms operate and ultimately how they co-operate with each other.

! Determine all possible pair-wise interactions for every S. cerevisiae protein.

The third global scale analysis we will consider is the systematic determination of protein-protein interactions in S. cerevisiae. This essentially involves a systematic test of all pair-wise combinations between all 5,800 yeast proteins. Individual matings to test >33 million combinations isn’t feasible, so mating pools of 100 strains in all combinations has become the preferred approach. Only the diploid strains where the Gal4 DB-fusion and the Gal4 AD-fusion proteins interact will be able to grow on galactose medium without uracil and histidine, as well as turning blue when grown on galactose and X-gal. The plasmids present in such diploids are then sequenced to determine which proteins are fused to the Gal4 AD and DB domains. This systematic approach to cataloguing all possible protein-protein interactions for yeast proteins yielded many more interactions than originally thought. Admittedly the yeast two hybrid is

Gal4 chimeric proteins representing all 5,800 proteins fused to the Gal4 Activation Domain and to the DNA

Binding domain.

Gal4 DNA Binding Domain One of 5,800 proteins

Gal4 Activation DomainOne of 5,800 proteins

5,800 Mat" yeast strains

5,800 Mata yeast strains

Gal4 chimeric proteins representing all 5,800 proteins fused to the Gal4 Activation Domain and to the DNA

Binding domain.

Gal4 DNA Binding Domain One of 5,800 proteins

Gal4 Activation DomainOne of 5,800 proteins Gal4 Activation DomainGal4 Activation DomainOne of 5,800 proteins

5,800 Mat" yeast strains

5,800 Mata yeast strains

X ZX Z

LacZ, URA3, HIS3

Select for diploids that can grow in the absence of Uracil and Histidine and which are blue on X-gal

Individual strains

5,800 Mat"#strains X 5,800 Mata strains 33,640,000 matings

Pools of 100 strains

58 pools Mat" strains X 58 pools Mata strains 3,364 matings

X ZX Z

LacZ, URA3, HIS3

Select for diploids that can grow in the absence of Uracil and Histidine and which are blue on X-gal

LacZ, URA3, HIS3

Select for diploids that can grow in the absence of Uracil and Histidine and which are blue on X-gal

Individual strains

5,800 Mat"#strains X 5,800 Mata strains 33,640,000 matings

Pools of 100 strains

58 pools Mat" strains X 58 pools Mata strains 3,364 matings

7 Fall 2006 – 7.03

quite noisy, giving many false positive interactions, but even so alternative methods (that we do not have time to consider in detail) have confirmed many of these interactions. When all of the known protein-protein interaction data is assembled, we see the surprising fact that > 5,000 proteins can be connected

together by > 14,000 protein interactions in a continuous web. Indeed, the interaction data for Gal4 embedded within this web makes sense and adds some new information. Such “Interactomes” are being developed for all the usual organisms, and the C. elegans interactome is particularly well developed. One of the major revelations has been that proteins from pathways that were previously thought to be totally unconnected, turn out to have interacting proteins.

Embedded in this complex web of interactions we can find those proteins that bind Gal4

Gal 80

Gal 4

Gal 3

Gal 1

Gal 11

• Gal1 can pinch-hit for Gal:

• Gal11 turns out to be a subunit of the PolII

transcription machinery so Gal4 communicates with

PolII Via Gal11

Embedded in this complex web of interactions we can find those proteins that bind Gal4

Gal 80Gal 80

Gal 4Gal 4

Gal 3Gal 3

Gal 1Gal 1

Gal 11Gal 11

• Gal1 can pinch-hit for Gal:

• Gal11 turns out to be a subunit of the PolII

transcription machinery so Gal4 communicates with

PolII Via Gal11